去创作
rllab 项目中求 KL 散度对应的代码
使用时可去掉参数 self，当成独立的函数。注意代码中使用了 theano 库，使用 tensorflow 和 numpy 时相应替换即可。
def kl_sym(self, old_dist_info_vars, new_dist_info_vars):
old_means = old_dist_info_vars["mean"]
old_log_stds = old_dist_info_vars["log_std"]
new_means = new_dist_info_vars["mean"]
new_log_stds = new_d ... ...
